Actual Needs Criteria for Assessing Data Classification Platforms

Abstract

A Software tools have an important role in different research areas. Generally they provide time and efforts saving. In computer science filed these tools can help in communications, web site development, software metrics finding, data mining, machine learning and many other fields. There are many specialized tools built to support specific purpose. Users and researchers spend a lot of time and efforts to select between the large amounts of the available platforms. Each has its own characteristics, some are open source and the other licensed with trial version to test them. In this work we will focus on some platforms related to data mining research area. The selected tools represent widely used and trusted ones with most updated version. We will study platforms from different perspectives. They have different data processing features, but they support common algorithms helps us to evaluate between them. Four data mining tools and four data set were selected. The assessment procedure done from multi-points of view as we will see in the methodology section of this article. The criteria collected from a survey done among a population of researchers interested in the field of data mining and machine learning.                The Contribution of this work is to assess the selected platforms depending on new actual needs criteria. These criteria give a clear idea for the researchers to determine the best platform according to their resources. The results highlighted the power for each platform. Orange and Weka show best performance over the rest. These results will be the guide for beginners or researchers out the computer science field to select the appropriate platform for their needs and available resources.